MA
AI News: 6 New Models & A Nano Banana KillerMatt Wolfe16 Oct 2026
Decision models, Nimble-v4 and v4-max, that support multiple choice, yes/no and rubric scores. Claimed to outperform Jev on shared benchmarks.
NI
Nimble-v4
by Bespoke Labs
Decision models from Bespoke Labs that output multiple-choice, yes/no and rubric-score judgments.
Updated EU status checked Last shown
Nimble-v4 at a glance
- What is Nimble-v4?
- Nimble-v4 is a model from Bespoke Labs. Decision models from Bespoke Labs that output multiple-choice, yes/no and rubric-score judgments.
- Who makes Nimble-v4?
- Nimble-v4 is made by Bespoke Labs.
- Does Nimble-v4 work in the EU?
- Unverified. No explicit evidence found. The vendor's privacy policy has an EEA/UK section and describes international transfer safeguards, but nothing specific to Nimble-v4 availability. Last verified 14 Nov 2026.
- How much does Nimble-v4 cost?
- Pricing unknown. No pricing information found.
- Where was Nimble-v4 shown?
- In 1 creator video by Matt Wolfe. Most recently by Matt Wolfe on 16 Oct 2026, at 23:00 in “AI News: 6 New Models & A Nano Banana Killer”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
About
Nimble-v4 and Nimble-v4-max are "decision" models from Bespoke Labs, a San Francisco Bay Area applied AI research lab. According to a YouTube news roundup (Matt Wolfe), they support multiple-choice, yes/no and rubric-score outputs, and are claimed to outperform a competing model ("Jev") on shared benchmarks. That claim comes from a creator summary and was not verified against vendor documentation.
The official Bespoke Labs site provided here does not describe Nimble-v4 at all. It focuses on RL environments, agent evaluation and optimization (the GEPA optimizer), and benchmarks such as Terminal-Bench and AutoResearchExam. Details such as parameter count, context length, pricing, weights availability and license are not documented in the available material.
Likely uses are judging or scoring tasks, such as evaluation, grading against rubrics and classification-style decisions, though the vendor has not confirmed this in the provided pages.
Key features
- Multiple-choice decisions
- Yes/no decisions
- Rubric-score outputs
- Two variants: Nimble-v4 and Nimble-v4-max
1 moment
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Keep exploring
Similar AI tools
Closest neighbours by meaning, not just by category.
BN
Bespoke Nimble 9B
Model · Bespoke Labs
Apache-2.0 LoRA adapter on Qwen3.5-9B that scores answer choices, booleans and rubric levels against a given context.
Available in EULanguage Models
18 Sept 2026Open source
PA
PaperBench
Research · OpenAI
OpenAI benchmark testing whether AI agents can replicate ICML 2024 papers from scratch, graded by hierarchical rubrics.
Available in EUResearch
2 Apr 2025Open source
AA
Agents A1
Model
35B Mixture-of-Experts open-weights model from InternScience for long-horizon agentic search, engineering and tool use.
Available in EULanguage Models
Release date n/aOpen source
HY
Hy3
Model · Tencent
Tencent's open-weights 295B Mixture-of-Experts model (21B active) for reasoning and agentic coding.
UnverifiedLanguage Models
Release date n/aOpen source
SA
Sage A1.5
Model · Alpha Innovator
35B mixture-of-experts open-weights agentic model for engineering, long-horizon search and scientific research.
Available in EULanguage Models
Release date n/aOpen source
ML
Mistral Large 4
Model · Mistral
Mistral's roughly 1-trillion-parameter multimodal mixture-of-experts model with open weights, trained in Europe.
Available in EULanguage Models
Release date n/aAPI